Chatbot Based On Knowledge Base In Automotive Field

Since 1950,man-machine conversation,one of the major research directions in natural language processing,has been the topic of research and exploration of predecessors continuously.Based on the simulation of expression process of human natural language,man-machine dialogue system is created to make proper conversations between human and machine.Despite of the great progress in the realm of human-computer interaction,there still exist some problems in the field.It has been observed that grasping the necessary background knowledge prior to the conversation is key to simulating the human way of conversation with the system.The content of conversation can only be more fluent and natural when we have had the relevant background knowledge.Therefore,it is also the core research content of this paper to integrate the background knowledge of conversation into the man-machine conversation model.According to the functions of chatbots,they can be divided into two major categories below: One type of chatbots is task-oriented chatbots,which are required of high accuracy in their answers and large and accurate knowledge databases.What is more,many pre-set templates are introduced for them to answer the questions simply to ensure the accuracy of answer-giving.Although the accuracy is enhanced,the flexibility of the model is greatly lowered,which justifies why it is mainly applied to simple and repetitive information exchange.The other type of chatbots is the non-task-based human-computer conversation system,which is characterized with less purposeful and more intelligent responses during the interactions with humans.However,non-task-based chatbots have some disadvantages such as monotonous reply content and lack of valuable information for users.In view of the deficiency of conventional task-based chatbots mentioned above,the paper carries on an in-depth study,proposing that the richness of the model-generated dialogue can be improved by incorporating the background knowledge into the traditional model through coding.In the meantime,it is also pointed out that the problem of inconsistent personality ofchatbot in multiple rounds of conversation can be eased to some extent with the current knowledge to fix the topic of conversation.Finally,an Android-based instant messaging App is implemented in the paper.By encapsulating the afore-said model in the paper into a chatbot in the App,it enables the communication with users in the field of automobile as well as simple conversations with users in daily life.